RESEARCH ARTICLE
Psychometric properties of the Brisbane Burn Scar Impact Profile in adults with burn scars Zephanie Tyack1,2*, Roy Kimble1, Steven McPhail2,3, Anita Plaza4, Megan Simons1,5 1 Centre for Children’s Burns and Trauma Research, Children’s Health Research Centre, The University of Queensland, South Brisbane, Queensland, Australia, 2 Institute of Health and Biomedical Innovation and School of Public Health and Social Work, Queensland University of Technology, Kelvin Grove, Queensland, Australia, 3 Centre for Functioning and Health Research, Metro South Health, Buranda, Australia, 4 Royal Brisbane and Women’s Hospital, Herston, Queensland, Australia, 5 Department of Occupational Therapy, Lady Cilento Children’s Hospital, South Brisbane, Queensland, Australia
a1111111111 a1111111111 a1111111111 a1111111111 a1111111111
* [email protected]
Abstract Objective
OPEN ACCESS Citation: Tyack Z, Kimble R, McPhail S, Plaza A, Simons M (2017) Psychometric properties of the Brisbane Burn Scar Impact Profile in adults with burn scars. PLoS ONE 12(9): e0184452. https:// doi.org/10.1371/journal.pone.0184452 Editor: Leila Harhaus, BG Trauma Center Ludwigshafen, GERMANY Received: March 9, 2017 Accepted: August 24, 2017 Published: September 13, 2017 Copyright: © 2017 Tyack et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: The individual data used in this study cannot be made publicly available due to ethical constraints and governance arrangements to protect patient privacy and ensure appropriate data use. Data can be requested from the data custodians with appropriate ethical and governance approvals from Metro South Health who can be contacted at the following email: [email protected]. Interested readers can address data requests to: z.tyack@uq. edu.au.
The aim of the study was to determine the longitudinal validity, reproducibility, responsiveness and interpretability of the adult version of the Brisbane Burn Scar Impact Profile, a patient-report measure of health-related quality of life.
Methods A prospective longitudinal cohort study of patients with or at risk of burn scarring was conducted at three assessment points (at baseline around the time of wound healing, one to two weeks post-baseline and 1-month post-baseline). Participants attending a major metropolitan adult burn centre at baseline were recruited. Participants completed the Brisbane Burn Scar Impact Profile and the 36-item Short Form Health Survey and Patient Observer Scar Assessment Scale. Intraclass Correlation Coefficients (ICCs), smallest detectable change, percentage of those who improved, stayed the same or worsened and Area under the Receiver Operating Characteristic Curve (AUC) were used to test the aim.
Results Data were included for 118 participants at baseline, 68 participants at one to two weeks and 57 participants at 1-month post-baseline. All groups of items had acceptable reproducibility, except for the overall impact of burn scars (ICC = 0.69), the impact of sensations which was not expected to be stable (ICC = 0.63), mobility and daily activities (ICC = 0.63, 0.67 respectively). The responsiveness of six out of seven groups of items able to be tested against external criterion was supported (AUC = 0.72–0.75). Hypothesised correlations of changes in the Brisbane Burn Scar Impact Profile items with changes in criterion measures generally supported longitudinal validity (e.g., nine out of thirteen hypotheses using the SF-36 as an external criterion were supported). Internal consistency estimates, item-total and inter-item correlations indicated there was likely redundancy of some groups of items, particularly in
PLOS ONE | https://doi.org/10.1371/journal.pone.0184452 September 13, 2017
1 / 21
Brisbane Burn Scar Impact Profile
Funding: Queensland Children’s Medical Research Institute provided funding to support the use of consumables and time for the lead investigator to collect the data. Central Queensland Hospital and Health Service provided staff support for the lead investigator to oversee the project. The Queensland Health, Health Practitioner Research Scheme 2012–2013 provided funding for a research assistant to assist to collect the data. The Australian Catholic University provided staff support for the lead investigator to collect data. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing interests: RK was a member of the Queensland Children’s Medical Research Centre at the time that funding was provided for the study. ZT was a staff member of the Central Queensland Hospital and Health Service District and Australian Catholic University at the time those organisations provided support for the study. AP was staff member at the Royal Brisbane and Women’s Hospital where the research was conducted. This does not alter our adherence to PLOS ONE policies on sharing data and materials.
the relationships and social interaction, appearance and emotional reactions items (Chronbach’s alpha range = 0.94–0.95).
Conclusion Support was found for the reproducibility, longitudinal validity, responsiveness and interpretability of most groups of Brisbane Burn Scar Impact Profile items and some individual items in the test population. Potential redundancy of items should be investigated further.
Introduction Burns can have devastating physical and psychological effects on individuals. In addition, treatments for burns and burn-related scarring use substantial amounts of healthcare resources [1] which, in the case of scar-related interventions, can continue long after the acute burn management has been completed. Scarring is a normal part of healing after tissue damage; however, in some patients an exaggerated response occurs involving complex molecular and cellular processes that are thought to be linked to immune [2] and sensorineural responses. The most common form of these exaggerated scar responses are hypertrophic scars [3] that are characterised by redness and thickness. It is estimated that 32 to 72 percent of patients with burns will develop hypertrophic scars [4]. The substantial burden of disease attributed to burn scars is demonstrated by findings that those with burn scarring have a significantly higher risk of readmission and have treatment costs that are 5.6 times higher than those without scarring [5]. Studies of the effectiveness of scar interventions have been limited by the use of outcome measures with unknown psychometric properties among people with burn scarring and that have frequently not included the measurement of health-related quality of life. The importance of measuring health-related quality of life in patients with burn scarring has been highlighted. Studies have found associations between burn scar severity, visibility and health-related quality of life [6], and limitations in three or more domains of health-related quality of life have been found in adults up to and including 18 months post-burn [7]. A new condition-specific patient-report outcome measure (PROM) of health-related quality of life was developed for people at risk of, or with burn scarring, as no such measure existed (termed the Brisbane Burn Scar Impact Profile—BBSIP, available at www.ccbtr.com.au [8]. This is the first known burn scar specific PROM measuring health-related quality of life. It is also the first multidimensional health-related PROM for people with burns to be developed based on interviews of patients themselves (N = 30 adults and children) [8], as existing burn-specific multidimensional health-related PROMs were developed using a literature review and expert opinion [9]. Content validity is also supported by the item development being based on a burn-scar specific conceptual framework of health-related quality of life [10]. Items were grouped into overall impact of scars; the intensity, frequency and impact of itch, pain and other sensations; work and daily activities (mobility and daily activities items); relationships and social interaction; and appearance, emotional reactions; and physical symptoms. The measure included items distinct from other burn-specific PROMs for burns including items measuring the intensity of tightness; sensitivity to cold, touch, or clothing; the impact of sensations when going downhill or downstairs; the impact of scar treatments and fatigue linked to burn scars; and the impact of scars on daily routines [8]. Although preliminary content validation has been conducted [8] work was required to test other psychometric properties of the Brisbane Burn Scar Impact Profile (BBSIP). The aim of
PLOS ONE | https://doi.org/10.1371/journal.pone.0184452 September 13, 2017
2 / 21
Brisbane Burn Scar Impact Profile
this study was to test the longitudinal validity, reproducibility, and responsiveness of the BBSIP in the post-acute period in line with an evaluative purpose (tracking changes in scarring over this period).
Materials and methods Design, setting, participants, and interventions A prospective longitudinal cohort study of patients with three assessment points ((baseline around the time of wound healing, one to two weeks post-baseline (termed 1-week follow-up) and 1-month post-baseline (termed 1-month follow-up)) was conducted. The setting was an outpatient clinic that provided burn care to adults at a major metropolitan Australian hospital in Brisbane. Consecutive patients were sampled from 2013 to 2015 (17/04/2013–04/07/2013; 26/09/2013–10/10/2013; 16/01/2014–28/07/2014; 3/11/2014–24/11/2014) based on the availability of two assessors. A sample size of 50 participants was sought for testing longitudinal validity, reproducibility, and responsiveness which is considered an appropriate sample size for testing those psychometric properties [11]. This sample size was sufficient to detect an ICC of 0.80 with 95% confidence intervals from 0.70 to 0.90 [12]. A sample size of 49 participants was also sufficient to detect a one sample correlation of 0.39 with 80 percent power and an alpha of 0.05 [13]. Eligibility criteria were: patients with scarring or with the potential to scar who attended the treating centre for their acute burn or follow-up management, aged over 18 years, having completed the BBSIP on at least one occasion, having burn wounds that were at least approximately 85% healed with the potential to scar at the baseline assessment. Patients with a range of burn scar severity (including those with scarring that restricted range of motion) were invited to participate. Patients with a cognitive, intellectual or physical impairment that impaired communication or memory were excluded, as were those with scars not caused by a burn injury and those who had eye or respiratory burns alone. Participants received face-to-face usual care interventions during the study period which was tailored to their clinical presentation and the priorities identified by the treating team. Usual care in this setting included pressure garments, wound dressings, exercises, taping to prevent or manage contractures, oro-facial orthoses or splints to prevent or manage contractures, return-to-work programs, skin moisturizers, skin massage, medication for sleep, itch or psychopathology, psychological screening and non-pharmacological intervention such as counselling. The length and frequency of intervention received varied across participants. After skin healing outpatient appointments were typically scheduled weekly or fortnightly initially, if the patient required pressure garments that needed to be measured or fitted. Patients requiring ongoing scar management were typically reviewed monthly or second monthly if they lived locally or every three to six months if they lived a long distance from the treating centre (with regular follow-up continued locally). The majority of the participants received their acute burn care in the setting where the study was conducted.
Questionnaires and other outcome measures The adult version of the BBSIP, measuring health-related quality of life of adults with burns scars, was tested. The development of the measure, preliminary content validation and a conceptual framework linked to development of the measure have been previously described [8]. The measure consists of 66 items and 10 item groupings. Items of the BBSIP are rated using a range of response formats including dichotomous scales for items such as the presence or absence of open wounds; 11-point numeric rating scales for the intensity of sensory items (where 0 indicates the absence of the sensation and 10 indicates the sensation as bad as it can
PLOS ONE | https://doi.org/10.1371/journal.pone.0184452 September 13, 2017
3 / 21
Brisbane Burn Scar Impact Profile
possibly be); 7-point likert scales for overall impact, daily living activities, emotional reactions, and social functioning; and 5-point likert scales for the frequency of sensory items and intensity of physical symptoms. The lower end of each scale indicates worse health-related quality of life. Other measures included were the observer and patient scales of the Patient Observer Scar Assessment Scale (POSAS) and the 36-item Short Form Health Survey (SF-36, version 2.0). The POSAS has been deemed to have acceptable reliability but indeterminant content and construct validity and internal consistency in a 2012 systematic review of burn scar rating scales [14]. Since that review, further work has supported the internal consistency and predictive validity of the patient and observer scale [15]. Structural validity has generally been supported using RASCH analysis although inclusion of the surface area items when assessing burn scars was not supported [16] and thus was not included in this study. The SF-36 is a generic measure of health-related quality of life that consists of eight subscales that were used in this study (physical functioning, role-physical, bodily pain, general health, vitality, social functioning, role emotional and mental health). The eight SF-36 dimensions were scored on 0 to 100 percentage scales with lower values indicating worse health-related quality of life. The SF-36 has been validated in adult patients with burns with discriminant and temporal validity supported [17]. Demographic details and clinical characteristics were collected from patients or their medical records and included gender, highest level of education attained, scar location, restrictions in range of motion (or joint contractures or scars pulling on other body parts), skin grafting, age, Fitzpatrick skin type, percent total body surface area burned (TBSA), days post-burn, and days to wound healing.
Procedures The COnsensus based Standards for the selection of health Measurement INstruments (COSMIN) checklist [18] was used as a guide for reporting the psychometric properties. All outcome measures were completed in paper format at baseline, 1-week follow-up (to determine reproducibility) and 1-month follow-up (to determine longitudinal validity and responsiveness). The criteria of 85% skin healing for baseline measurements to be taken was judged visually by an assessor. Two assessors were involved in obtaining consent, providing instructions and the PROMs to participants. The order of the PROMs and instructions were standardised. A 1- to 2-week test retest interval was chosen as the period during which physical scar properties such as thickness and vascularity were expected to be relatively stable, based on other studies that have examined reproducibility of scar measures using this interval [19,20]. The POSAS and scar intensity components of the BBSIP were administered for the worst area of scarring, which patients were asked to identify prior to the measures being administered. The only exception to this was when the scar site was difficult to relocate. In this case another worst area of scarring was chosen by the patient. This worst area was identified using a 3 by 3cm area, marked on the skin using a soft tipped skin pencil. When patients were unable to return to the hospital to complete the measures, the 1-week follow-up and 1-month follow-up measures were posted with a self-addressed return envelope and instructions for completion at home. Approximate time to wound healing was obtained from the medical records or judged visually by the assessor when patients returned to the hospital.
Statistical analysis Validity (convergent and divergent). Hypothesised correlations between scores on the BBSIP items and total score and respective items and subscale scores of the POSAS and SF-36 were specified a-priori (as highlighted in the tables) and were expected to be higher than 0.3 at
PLOS ONE | https://doi.org/10.1371/journal.pone.0184452 September 13, 2017
4 / 21
Brisbane Burn Scar Impact Profile
baseline and higher than correlations with other items at 1-month follow-up. Correlations of changes in BBSIP scores with changes in respective SF-36 and POSAS items over the 1-month follow-up were also analysed to determine longitudinal validity. In brief, the pain item of the POSAS was expected to correlate more strongly with the pain item of the BBSIP as well as with the tightness item based on our previous work. The patient’s overall opinion of the scar on the POSAS was expected to correlate strongly with the appearance items of the BBSIP at baseline and for change from baseline to 1-month follow-up compared to respective correlations with other items. The sensory-related items of the BBSIP were expected to correlate more strongly with the itch and pain items of the POSAS than with other POSAS items, with correlations between respective items expected to be strongest (i.e. itch on the BBSIP with itch on the POSAS). Changes in BBSIP individual and group items related to mood and emotional responses were expected to correlate more strongly with the SF-36 mental health subscale than with other subscales. BBSIP individual and group items related to ADL, mobility, work and daily activities, and physical scar symptoms were expected to correlate more strongly with SF36 physical functioning subscales than with other SF-36 subscales. BBSIP individual and group items related to social functioning were expected to correlate more strongly with the SF-36 social functioning subscale than with other SF-36 subscales. The BBSIP items of tight scars making you tired was expected to correlate more strongly with the vitality subscale of the SF36 than with other SF-36 subscales. Descriptive statistics, internal consistency and inter-item correlations. Descriptive statistics were used to describe the study sample (e.g. means and standard deviations, medians and interquartile ranges for continuous data, and frequencies and percentages for count data). Differences in demographic and clinical variables of participants between the time points were examined using related-sample data analyses (Wilcoxon Signed Rank Test for ordinal data and McNemar Test for nominal data). Missing data on the SF-36 data were imputed using the mean values of the remaining dimension items as recommended [21], when less than 50% of the items were missing. When 50% or greater of SF-36 subscale items were missing, data were treated as missing. Individual BBSIP items scored as not applicable were treated as missing and only complete data were used in individual item analyses. Total BBSIP subscale scores were calculated where the number of missing or not applicable items was less than 50% of the subscale items, by summing the available scores divided by the number of available items. Chronbach’s alpha and inter-item and item-total correlations using Spearman’s rho were used to indicate the internal consistency and inter-relatedness of items considered to group into sub-scales. Chronbach’s alpha of 0.7 to 0.9, item-total correlations of 0.3 to 0.7 [22] and interitem correlations of approximately 0.20 to 0.40 [23] were considered ideal with higher values indicating potential redundancy of items. Reproducibility and responsiveness. Reproducibility was tested by examining agreement and reliability. Agreement between baseline and 1-week follow-up was examined for all items using the number and percentage of items with exact agreement, agreement within 1-point and agreement within 2-points were also calculated for items with three or more response options. Agreement was also examined using standard errors of measurement (SEM), and smallest detectable change (SDC), where assumptions of normality were met. The SEM was p calculated as σ2 (where σ2 was the mean square error term from the ICC ANOVA) [11, 24] p and the SDC was calculated as 1.96 x 2 x SEM. Reliability was examined using Cohen’s kappa for the dichotomous item and Intraclass Correlation Coefficients (ICC’s) for other items. ICC’s were calculated using a two-way, random effects model and associated confidence intervals and absolute agreement, with an ICC of greater than 0.7 considered acceptable [11]. Responsiveness was determined using Receiver Operating Characteristic Curves (ROC) to assess the area under the curve (AUC) corresponding to the correct identification of patients
PLOS ONE | https://doi.org/10.1371/journal.pone.0184452 September 13, 2017
5 / 21
Brisbane Burn Scar Impact Profile
who changed using external criterion. The AUC can range from 0.5 (no discriminatory accuracy) to 1.0 (perfect accuracy) with an AUC of 0.7 indicating acceptable responsiveness of the item or domain [11]. Spearman’s rho was used to examine correlations between change scores and change in the external criterion with a correlation of 0.3 required for the external criterion to establish responsiveness [25]. Two comparisons were used where the POSAS patient scale items were an appropriate external criterion (worsening verses no worsening) and three comparisons were used where the SF-36 subscales were an appropriate external criterion (change versus no change), where change was defined as greater than the mean change per month on SF-36 subscales found in a previous study of an adult burn population [17] (e.g. greater than 5 points on the physical functioning scale versus less than or equal to 5 points). Interpretability. It was assumed that evidence of some improvement would be detected in individual items and groups of items in the presence of tailored scar and psychosocial interventions that were typically provided after the baseline measurement, thus the percentage of those who stayed the same, improved and worsened was determined for the 1-month followup. At 1-month follow-up a greater number of participants were expected to have scar thickness and roughness stay the same or worsen than improve but for scar colour and sensory symptoms a larger number were expected to stay the same or improve compared to worsen, based on previous work on the trajectory of patient-reported scar severity in people receiving scar interventions similar to those provided to the study participants after burns [15]. Interpretability was also examined using the medians and interquartile ranges of subgroups expected to differ and percentage of missing items. Three subgroups of improved, stable and deteriorated were examined using SF-36 physical, social and mental health subscales and POSAS patient scale items of pain and overall opinion as criterion, with theory-based cutoffs used for the SF-36. Both single items and groups of items of the BBSIP were tested where appropriate. Significance was reported using p-vales of less than 0.05. Statistical analyses were performed using IBM SPSS Statistics for Windows, Version 23.0. (Armonk, New York: IBM Corporation). Ethical approval was provided by Metro South Human Research Ethics Committee in 2011 (HREC12/QPAH/595) with written informed consent to participate obtained from all participants included in the study.
Results The number of participants who were included in the study and completed testing at each timepoint is presented in Fig 1. Participant sociodemographic, injury, scar and skin type characteristics for the baseline, 1-week and 1-month follow-up samples are reported in Table 1. Participants were predominantly male, had a median age of 34 years, and the majority had a secondary education and received skin grafting. There were no significant differences in the characteristics of participants between baseline and 1-week follow-up and baseline and 1-month follow-up (Table 1). At baseline the mean scar thickness score reported by an observer using the POSAS was 3.68 (SD = 2.00) which remained the same at 3.68 (SD = 2.03) at 1-month follow-up (scale range 1 to 10). At baseline the mean scar thickness reported by the patient using the POSAS was higher (mean = 5.82, SD = 2.77) and also remained relatively constant at 1-month followup (mean = 5.57, SD = 2.91). At baseline the mean overall opinion of the patient regarding their scar using the POSAS was 6.97 (SD = 2.33) which reduced slightly to 6.39 (SD = 2.43) at 1-month follow-up. The percentage of missing scores for individual items varied from zero to three percent across the items at baseline and from zero to five percent across the items at 1-week follow-up
PLOS ONE | https://doi.org/10.1371/journal.pone.0184452 September 13, 2017
6 / 21
Brisbane Burn Scar Impact Profile
Fig 1. Flowchart of participants included and excluded. https://doi.org/10.1371/journal.pone.0184452.g001
(with the highest percent missing items on overall impact on life (5%) and driving a car or vehicle (3%)). At 1-month follow-up the percentage of missing items varied from zero to four percent across the items, with the highest percentage of missing items for appearance, emotional reactions and physical symptom items. Cronbach’s alpha estimates as well as item-total and inter-item correlations (S1 Table) generally supported the internal consistency of subscales, but indicated there may be some redundancy of items within some item groupings, particularly for the relationships and social interaction items, appearance items and emotional reactions items. Generally, hypothesised correlations of change in BBSIP items and groups of items with changes in criterion measures supported longitudinal validity (12 out of 19 hypotheses using the POSAS as the external criterion were supported and 9 out of 13 hypotheses using the SF-36 as an external criterion were supported). Correlations at baseline and between changes from baseline to 1-month follow-up on the BBSIP items and relevant POSAS patient scale items (Table 2) supported the validity and longitudinal validity of individual items of the overall impact of itch, pain and other sensations on your life, itch intensity, tightness intensity, sensitivity intensity, pain intensity, and tight and thick scar items; and the group items of the total sensory intensity score and physical symptoms total score. Correlations between changes from baseline to 1-month follow-up on the BBSIP items and relevant SF-36 dimensions (Table 3) supported the validity and longitudinal validity of the individual items of the overall impact on
PLOS ONE | https://doi.org/10.1371/journal.pone.0184452 September 13, 2017
7 / 21
Brisbane Burn Scar Impact Profile
Table 1. Participant sociodemographic, injury, scar and skin type characteristics at each testing timepoint. Number of participants (% of sample)a
Participant characteristics
Baseline
1-week follow-upb
1-mth follow-upb
Number of participants (scars)
118 (118)
65 (65)
57 (57)
Male gender
87 (74%)
49 (75%)
42 (74%)
(p = .32)c
(p = 1.00)c
Education—highest level attained Bachelor degree of higher (1)
15 (13%)
8 (12%)
4 (7%)
Diploma, Advanced diploma or post school certificate (2,3)
32 (27%)
20 (31%)
18 (32%)
Secondary education (4–7)
53 (45%)
29 (45%)
26 (46%)
Primary education or lower (8)
5 (4%)
3 (5%)
3 (5%)
Other (9,10)
2 (2%)
2 (3%)
2 (3%)
Missing
7 (6%)
2 (3%)
4 (7%)
Upper limbs
64 (54%)
35 (54%) (p = 1.00)
24 (42%) (p = 1.00)
Lower limbs
64 (54%)
23 (35%) (p = 1.00)
35 (61%) (p = 1.00)
Torso
30 (25%)
16 (25%) (p = 1.00)
18 (32%) (p = 1.00)
Head and neck
22 (19%)
13 (20%) (p = 1.00)
15 (26%) (p = 1.00)
45 (38%)
24 (37%) (p = 1.00)
24 (42%) (p = 1.00)
Scar location
Number of participants with restriction in range of motion or joint contractures or scars pulling on other body parts due to scarring Grafted
73 (62%)
45 (69%) (p = 1.00)
41 (72%) (p = 1.00)
3 (3%)
1 (2%)
0 (0%)
1 (1), 0–16
1 (1), 0–16 (p = 1.00)
1 (2), 0–16 (p = 1.00)
3 (3%)
1 (2%)
0 (%)
Median age in years (interquartile range), range
34 (21), 18–85
34 (26), 18–80 (p = .32)
34 (22), 19–80 (p = 1.00)
Median Fitzpatrick skin type (interquartile range), range
3 (3–4), 1–5
3 (3–4), 1–5 (p = 0.32)
4 (3–4), 1–5 (p = 1.00)
1 (1%)
0 (0%)
0 (0%)
4 (1–10),
Psychometric properties of the Brisbane Burn Scar Impact Profile in adults with burn scars Zephanie Tyack1,2*, Roy Kimble1, Steven McPhail2,3, Anita Plaza4, Megan Simons1,5 1 Centre for Children’s Burns and Trauma Research, Children’s Health Research Centre, The University of Queensland, South Brisbane, Queensland, Australia, 2 Institute of Health and Biomedical Innovation and School of Public Health and Social Work, Queensland University of Technology, Kelvin Grove, Queensland, Australia, 3 Centre for Functioning and Health Research, Metro South Health, Buranda, Australia, 4 Royal Brisbane and Women’s Hospital, Herston, Queensland, Australia, 5 Department of Occupational Therapy, Lady Cilento Children’s Hospital, South Brisbane, Queensland, Australia
a1111111111 a1111111111 a1111111111 a1111111111 a1111111111
* [email protected]
Abstract Objective
OPEN ACCESS Citation: Tyack Z, Kimble R, McPhail S, Plaza A, Simons M (2017) Psychometric properties of the Brisbane Burn Scar Impact Profile in adults with burn scars. PLoS ONE 12(9): e0184452. https:// doi.org/10.1371/journal.pone.0184452 Editor: Leila Harhaus, BG Trauma Center Ludwigshafen, GERMANY Received: March 9, 2017 Accepted: August 24, 2017 Published: September 13, 2017 Copyright: © 2017 Tyack et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: The individual data used in this study cannot be made publicly available due to ethical constraints and governance arrangements to protect patient privacy and ensure appropriate data use. Data can be requested from the data custodians with appropriate ethical and governance approvals from Metro South Health who can be contacted at the following email: [email protected]. Interested readers can address data requests to: z.tyack@uq. edu.au.
The aim of the study was to determine the longitudinal validity, reproducibility, responsiveness and interpretability of the adult version of the Brisbane Burn Scar Impact Profile, a patient-report measure of health-related quality of life.
Methods A prospective longitudinal cohort study of patients with or at risk of burn scarring was conducted at three assessment points (at baseline around the time of wound healing, one to two weeks post-baseline and 1-month post-baseline). Participants attending a major metropolitan adult burn centre at baseline were recruited. Participants completed the Brisbane Burn Scar Impact Profile and the 36-item Short Form Health Survey and Patient Observer Scar Assessment Scale. Intraclass Correlation Coefficients (ICCs), smallest detectable change, percentage of those who improved, stayed the same or worsened and Area under the Receiver Operating Characteristic Curve (AUC) were used to test the aim.
Results Data were included for 118 participants at baseline, 68 participants at one to two weeks and 57 participants at 1-month post-baseline. All groups of items had acceptable reproducibility, except for the overall impact of burn scars (ICC = 0.69), the impact of sensations which was not expected to be stable (ICC = 0.63), mobility and daily activities (ICC = 0.63, 0.67 respectively). The responsiveness of six out of seven groups of items able to be tested against external criterion was supported (AUC = 0.72–0.75). Hypothesised correlations of changes in the Brisbane Burn Scar Impact Profile items with changes in criterion measures generally supported longitudinal validity (e.g., nine out of thirteen hypotheses using the SF-36 as an external criterion were supported). Internal consistency estimates, item-total and inter-item correlations indicated there was likely redundancy of some groups of items, particularly in
PLOS ONE | https://doi.org/10.1371/journal.pone.0184452 September 13, 2017
1 / 21
Brisbane Burn Scar Impact Profile
Funding: Queensland Children’s Medical Research Institute provided funding to support the use of consumables and time for the lead investigator to collect the data. Central Queensland Hospital and Health Service provided staff support for the lead investigator to oversee the project. The Queensland Health, Health Practitioner Research Scheme 2012–2013 provided funding for a research assistant to assist to collect the data. The Australian Catholic University provided staff support for the lead investigator to collect data. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing interests: RK was a member of the Queensland Children’s Medical Research Centre at the time that funding was provided for the study. ZT was a staff member of the Central Queensland Hospital and Health Service District and Australian Catholic University at the time those organisations provided support for the study. AP was staff member at the Royal Brisbane and Women’s Hospital where the research was conducted. This does not alter our adherence to PLOS ONE policies on sharing data and materials.
the relationships and social interaction, appearance and emotional reactions items (Chronbach’s alpha range = 0.94–0.95).
Conclusion Support was found for the reproducibility, longitudinal validity, responsiveness and interpretability of most groups of Brisbane Burn Scar Impact Profile items and some individual items in the test population. Potential redundancy of items should be investigated further.
Introduction Burns can have devastating physical and psychological effects on individuals. In addition, treatments for burns and burn-related scarring use substantial amounts of healthcare resources [1] which, in the case of scar-related interventions, can continue long after the acute burn management has been completed. Scarring is a normal part of healing after tissue damage; however, in some patients an exaggerated response occurs involving complex molecular and cellular processes that are thought to be linked to immune [2] and sensorineural responses. The most common form of these exaggerated scar responses are hypertrophic scars [3] that are characterised by redness and thickness. It is estimated that 32 to 72 percent of patients with burns will develop hypertrophic scars [4]. The substantial burden of disease attributed to burn scars is demonstrated by findings that those with burn scarring have a significantly higher risk of readmission and have treatment costs that are 5.6 times higher than those without scarring [5]. Studies of the effectiveness of scar interventions have been limited by the use of outcome measures with unknown psychometric properties among people with burn scarring and that have frequently not included the measurement of health-related quality of life. The importance of measuring health-related quality of life in patients with burn scarring has been highlighted. Studies have found associations between burn scar severity, visibility and health-related quality of life [6], and limitations in three or more domains of health-related quality of life have been found in adults up to and including 18 months post-burn [7]. A new condition-specific patient-report outcome measure (PROM) of health-related quality of life was developed for people at risk of, or with burn scarring, as no such measure existed (termed the Brisbane Burn Scar Impact Profile—BBSIP, available at www.ccbtr.com.au [8]. This is the first known burn scar specific PROM measuring health-related quality of life. It is also the first multidimensional health-related PROM for people with burns to be developed based on interviews of patients themselves (N = 30 adults and children) [8], as existing burn-specific multidimensional health-related PROMs were developed using a literature review and expert opinion [9]. Content validity is also supported by the item development being based on a burn-scar specific conceptual framework of health-related quality of life [10]. Items were grouped into overall impact of scars; the intensity, frequency and impact of itch, pain and other sensations; work and daily activities (mobility and daily activities items); relationships and social interaction; and appearance, emotional reactions; and physical symptoms. The measure included items distinct from other burn-specific PROMs for burns including items measuring the intensity of tightness; sensitivity to cold, touch, or clothing; the impact of sensations when going downhill or downstairs; the impact of scar treatments and fatigue linked to burn scars; and the impact of scars on daily routines [8]. Although preliminary content validation has been conducted [8] work was required to test other psychometric properties of the Brisbane Burn Scar Impact Profile (BBSIP). The aim of
PLOS ONE | https://doi.org/10.1371/journal.pone.0184452 September 13, 2017
2 / 21
Brisbane Burn Scar Impact Profile
this study was to test the longitudinal validity, reproducibility, and responsiveness of the BBSIP in the post-acute period in line with an evaluative purpose (tracking changes in scarring over this period).
Materials and methods Design, setting, participants, and interventions A prospective longitudinal cohort study of patients with three assessment points ((baseline around the time of wound healing, one to two weeks post-baseline (termed 1-week follow-up) and 1-month post-baseline (termed 1-month follow-up)) was conducted. The setting was an outpatient clinic that provided burn care to adults at a major metropolitan Australian hospital in Brisbane. Consecutive patients were sampled from 2013 to 2015 (17/04/2013–04/07/2013; 26/09/2013–10/10/2013; 16/01/2014–28/07/2014; 3/11/2014–24/11/2014) based on the availability of two assessors. A sample size of 50 participants was sought for testing longitudinal validity, reproducibility, and responsiveness which is considered an appropriate sample size for testing those psychometric properties [11]. This sample size was sufficient to detect an ICC of 0.80 with 95% confidence intervals from 0.70 to 0.90 [12]. A sample size of 49 participants was also sufficient to detect a one sample correlation of 0.39 with 80 percent power and an alpha of 0.05 [13]. Eligibility criteria were: patients with scarring or with the potential to scar who attended the treating centre for their acute burn or follow-up management, aged over 18 years, having completed the BBSIP on at least one occasion, having burn wounds that were at least approximately 85% healed with the potential to scar at the baseline assessment. Patients with a range of burn scar severity (including those with scarring that restricted range of motion) were invited to participate. Patients with a cognitive, intellectual or physical impairment that impaired communication or memory were excluded, as were those with scars not caused by a burn injury and those who had eye or respiratory burns alone. Participants received face-to-face usual care interventions during the study period which was tailored to their clinical presentation and the priorities identified by the treating team. Usual care in this setting included pressure garments, wound dressings, exercises, taping to prevent or manage contractures, oro-facial orthoses or splints to prevent or manage contractures, return-to-work programs, skin moisturizers, skin massage, medication for sleep, itch or psychopathology, psychological screening and non-pharmacological intervention such as counselling. The length and frequency of intervention received varied across participants. After skin healing outpatient appointments were typically scheduled weekly or fortnightly initially, if the patient required pressure garments that needed to be measured or fitted. Patients requiring ongoing scar management were typically reviewed monthly or second monthly if they lived locally or every three to six months if they lived a long distance from the treating centre (with regular follow-up continued locally). The majority of the participants received their acute burn care in the setting where the study was conducted.
Questionnaires and other outcome measures The adult version of the BBSIP, measuring health-related quality of life of adults with burns scars, was tested. The development of the measure, preliminary content validation and a conceptual framework linked to development of the measure have been previously described [8]. The measure consists of 66 items and 10 item groupings. Items of the BBSIP are rated using a range of response formats including dichotomous scales for items such as the presence or absence of open wounds; 11-point numeric rating scales for the intensity of sensory items (where 0 indicates the absence of the sensation and 10 indicates the sensation as bad as it can
PLOS ONE | https://doi.org/10.1371/journal.pone.0184452 September 13, 2017
3 / 21
Brisbane Burn Scar Impact Profile
possibly be); 7-point likert scales for overall impact, daily living activities, emotional reactions, and social functioning; and 5-point likert scales for the frequency of sensory items and intensity of physical symptoms. The lower end of each scale indicates worse health-related quality of life. Other measures included were the observer and patient scales of the Patient Observer Scar Assessment Scale (POSAS) and the 36-item Short Form Health Survey (SF-36, version 2.0). The POSAS has been deemed to have acceptable reliability but indeterminant content and construct validity and internal consistency in a 2012 systematic review of burn scar rating scales [14]. Since that review, further work has supported the internal consistency and predictive validity of the patient and observer scale [15]. Structural validity has generally been supported using RASCH analysis although inclusion of the surface area items when assessing burn scars was not supported [16] and thus was not included in this study. The SF-36 is a generic measure of health-related quality of life that consists of eight subscales that were used in this study (physical functioning, role-physical, bodily pain, general health, vitality, social functioning, role emotional and mental health). The eight SF-36 dimensions were scored on 0 to 100 percentage scales with lower values indicating worse health-related quality of life. The SF-36 has been validated in adult patients with burns with discriminant and temporal validity supported [17]. Demographic details and clinical characteristics were collected from patients or their medical records and included gender, highest level of education attained, scar location, restrictions in range of motion (or joint contractures or scars pulling on other body parts), skin grafting, age, Fitzpatrick skin type, percent total body surface area burned (TBSA), days post-burn, and days to wound healing.
Procedures The COnsensus based Standards for the selection of health Measurement INstruments (COSMIN) checklist [18] was used as a guide for reporting the psychometric properties. All outcome measures were completed in paper format at baseline, 1-week follow-up (to determine reproducibility) and 1-month follow-up (to determine longitudinal validity and responsiveness). The criteria of 85% skin healing for baseline measurements to be taken was judged visually by an assessor. Two assessors were involved in obtaining consent, providing instructions and the PROMs to participants. The order of the PROMs and instructions were standardised. A 1- to 2-week test retest interval was chosen as the period during which physical scar properties such as thickness and vascularity were expected to be relatively stable, based on other studies that have examined reproducibility of scar measures using this interval [19,20]. The POSAS and scar intensity components of the BBSIP were administered for the worst area of scarring, which patients were asked to identify prior to the measures being administered. The only exception to this was when the scar site was difficult to relocate. In this case another worst area of scarring was chosen by the patient. This worst area was identified using a 3 by 3cm area, marked on the skin using a soft tipped skin pencil. When patients were unable to return to the hospital to complete the measures, the 1-week follow-up and 1-month follow-up measures were posted with a self-addressed return envelope and instructions for completion at home. Approximate time to wound healing was obtained from the medical records or judged visually by the assessor when patients returned to the hospital.
Statistical analysis Validity (convergent and divergent). Hypothesised correlations between scores on the BBSIP items and total score and respective items and subscale scores of the POSAS and SF-36 were specified a-priori (as highlighted in the tables) and were expected to be higher than 0.3 at
PLOS ONE | https://doi.org/10.1371/journal.pone.0184452 September 13, 2017
4 / 21
Brisbane Burn Scar Impact Profile
baseline and higher than correlations with other items at 1-month follow-up. Correlations of changes in BBSIP scores with changes in respective SF-36 and POSAS items over the 1-month follow-up were also analysed to determine longitudinal validity. In brief, the pain item of the POSAS was expected to correlate more strongly with the pain item of the BBSIP as well as with the tightness item based on our previous work. The patient’s overall opinion of the scar on the POSAS was expected to correlate strongly with the appearance items of the BBSIP at baseline and for change from baseline to 1-month follow-up compared to respective correlations with other items. The sensory-related items of the BBSIP were expected to correlate more strongly with the itch and pain items of the POSAS than with other POSAS items, with correlations between respective items expected to be strongest (i.e. itch on the BBSIP with itch on the POSAS). Changes in BBSIP individual and group items related to mood and emotional responses were expected to correlate more strongly with the SF-36 mental health subscale than with other subscales. BBSIP individual and group items related to ADL, mobility, work and daily activities, and physical scar symptoms were expected to correlate more strongly with SF36 physical functioning subscales than with other SF-36 subscales. BBSIP individual and group items related to social functioning were expected to correlate more strongly with the SF-36 social functioning subscale than with other SF-36 subscales. The BBSIP items of tight scars making you tired was expected to correlate more strongly with the vitality subscale of the SF36 than with other SF-36 subscales. Descriptive statistics, internal consistency and inter-item correlations. Descriptive statistics were used to describe the study sample (e.g. means and standard deviations, medians and interquartile ranges for continuous data, and frequencies and percentages for count data). Differences in demographic and clinical variables of participants between the time points were examined using related-sample data analyses (Wilcoxon Signed Rank Test for ordinal data and McNemar Test for nominal data). Missing data on the SF-36 data were imputed using the mean values of the remaining dimension items as recommended [21], when less than 50% of the items were missing. When 50% or greater of SF-36 subscale items were missing, data were treated as missing. Individual BBSIP items scored as not applicable were treated as missing and only complete data were used in individual item analyses. Total BBSIP subscale scores were calculated where the number of missing or not applicable items was less than 50% of the subscale items, by summing the available scores divided by the number of available items. Chronbach’s alpha and inter-item and item-total correlations using Spearman’s rho were used to indicate the internal consistency and inter-relatedness of items considered to group into sub-scales. Chronbach’s alpha of 0.7 to 0.9, item-total correlations of 0.3 to 0.7 [22] and interitem correlations of approximately 0.20 to 0.40 [23] were considered ideal with higher values indicating potential redundancy of items. Reproducibility and responsiveness. Reproducibility was tested by examining agreement and reliability. Agreement between baseline and 1-week follow-up was examined for all items using the number and percentage of items with exact agreement, agreement within 1-point and agreement within 2-points were also calculated for items with three or more response options. Agreement was also examined using standard errors of measurement (SEM), and smallest detectable change (SDC), where assumptions of normality were met. The SEM was p calculated as σ2 (where σ2 was the mean square error term from the ICC ANOVA) [11, 24] p and the SDC was calculated as 1.96 x 2 x SEM. Reliability was examined using Cohen’s kappa for the dichotomous item and Intraclass Correlation Coefficients (ICC’s) for other items. ICC’s were calculated using a two-way, random effects model and associated confidence intervals and absolute agreement, with an ICC of greater than 0.7 considered acceptable [11]. Responsiveness was determined using Receiver Operating Characteristic Curves (ROC) to assess the area under the curve (AUC) corresponding to the correct identification of patients
PLOS ONE | https://doi.org/10.1371/journal.pone.0184452 September 13, 2017
5 / 21
Brisbane Burn Scar Impact Profile
who changed using external criterion. The AUC can range from 0.5 (no discriminatory accuracy) to 1.0 (perfect accuracy) with an AUC of 0.7 indicating acceptable responsiveness of the item or domain [11]. Spearman’s rho was used to examine correlations between change scores and change in the external criterion with a correlation of 0.3 required for the external criterion to establish responsiveness [25]. Two comparisons were used where the POSAS patient scale items were an appropriate external criterion (worsening verses no worsening) and three comparisons were used where the SF-36 subscales were an appropriate external criterion (change versus no change), where change was defined as greater than the mean change per month on SF-36 subscales found in a previous study of an adult burn population [17] (e.g. greater than 5 points on the physical functioning scale versus less than or equal to 5 points). Interpretability. It was assumed that evidence of some improvement would be detected in individual items and groups of items in the presence of tailored scar and psychosocial interventions that were typically provided after the baseline measurement, thus the percentage of those who stayed the same, improved and worsened was determined for the 1-month followup. At 1-month follow-up a greater number of participants were expected to have scar thickness and roughness stay the same or worsen than improve but for scar colour and sensory symptoms a larger number were expected to stay the same or improve compared to worsen, based on previous work on the trajectory of patient-reported scar severity in people receiving scar interventions similar to those provided to the study participants after burns [15]. Interpretability was also examined using the medians and interquartile ranges of subgroups expected to differ and percentage of missing items. Three subgroups of improved, stable and deteriorated were examined using SF-36 physical, social and mental health subscales and POSAS patient scale items of pain and overall opinion as criterion, with theory-based cutoffs used for the SF-36. Both single items and groups of items of the BBSIP were tested where appropriate. Significance was reported using p-vales of less than 0.05. Statistical analyses were performed using IBM SPSS Statistics for Windows, Version 23.0. (Armonk, New York: IBM Corporation). Ethical approval was provided by Metro South Human Research Ethics Committee in 2011 (HREC12/QPAH/595) with written informed consent to participate obtained from all participants included in the study.
Results The number of participants who were included in the study and completed testing at each timepoint is presented in Fig 1. Participant sociodemographic, injury, scar and skin type characteristics for the baseline, 1-week and 1-month follow-up samples are reported in Table 1. Participants were predominantly male, had a median age of 34 years, and the majority had a secondary education and received skin grafting. There were no significant differences in the characteristics of participants between baseline and 1-week follow-up and baseline and 1-month follow-up (Table 1). At baseline the mean scar thickness score reported by an observer using the POSAS was 3.68 (SD = 2.00) which remained the same at 3.68 (SD = 2.03) at 1-month follow-up (scale range 1 to 10). At baseline the mean scar thickness reported by the patient using the POSAS was higher (mean = 5.82, SD = 2.77) and also remained relatively constant at 1-month followup (mean = 5.57, SD = 2.91). At baseline the mean overall opinion of the patient regarding their scar using the POSAS was 6.97 (SD = 2.33) which reduced slightly to 6.39 (SD = 2.43) at 1-month follow-up. The percentage of missing scores for individual items varied from zero to three percent across the items at baseline and from zero to five percent across the items at 1-week follow-up
PLOS ONE | https://doi.org/10.1371/journal.pone.0184452 September 13, 2017
6 / 21
Brisbane Burn Scar Impact Profile
Fig 1. Flowchart of participants included and excluded. https://doi.org/10.1371/journal.pone.0184452.g001
(with the highest percent missing items on overall impact on life (5%) and driving a car or vehicle (3%)). At 1-month follow-up the percentage of missing items varied from zero to four percent across the items, with the highest percentage of missing items for appearance, emotional reactions and physical symptom items. Cronbach’s alpha estimates as well as item-total and inter-item correlations (S1 Table) generally supported the internal consistency of subscales, but indicated there may be some redundancy of items within some item groupings, particularly for the relationships and social interaction items, appearance items and emotional reactions items. Generally, hypothesised correlations of change in BBSIP items and groups of items with changes in criterion measures supported longitudinal validity (12 out of 19 hypotheses using the POSAS as the external criterion were supported and 9 out of 13 hypotheses using the SF-36 as an external criterion were supported). Correlations at baseline and between changes from baseline to 1-month follow-up on the BBSIP items and relevant POSAS patient scale items (Table 2) supported the validity and longitudinal validity of individual items of the overall impact of itch, pain and other sensations on your life, itch intensity, tightness intensity, sensitivity intensity, pain intensity, and tight and thick scar items; and the group items of the total sensory intensity score and physical symptoms total score. Correlations between changes from baseline to 1-month follow-up on the BBSIP items and relevant SF-36 dimensions (Table 3) supported the validity and longitudinal validity of the individual items of the overall impact on
PLOS ONE | https://doi.org/10.1371/journal.pone.0184452 September 13, 2017
7 / 21
Brisbane Burn Scar Impact Profile
Table 1. Participant sociodemographic, injury, scar and skin type characteristics at each testing timepoint. Number of participants (% of sample)a
Participant characteristics
Baseline
1-week follow-upb
1-mth follow-upb
Number of participants (scars)
118 (118)
65 (65)
57 (57)
Male gender
87 (74%)
49 (75%)
42 (74%)
(p = .32)c
(p = 1.00)c
Education—highest level attained Bachelor degree of higher (1)
15 (13%)
8 (12%)
4 (7%)
Diploma, Advanced diploma or post school certificate (2,3)
32 (27%)
20 (31%)
18 (32%)
Secondary education (4–7)
53 (45%)
29 (45%)
26 (46%)
Primary education or lower (8)
5 (4%)
3 (5%)
3 (5%)
Other (9,10)
2 (2%)
2 (3%)
2 (3%)
Missing
7 (6%)
2 (3%)
4 (7%)
Upper limbs
64 (54%)
35 (54%) (p = 1.00)
24 (42%) (p = 1.00)
Lower limbs
64 (54%)
23 (35%) (p = 1.00)
35 (61%) (p = 1.00)
Torso
30 (25%)
16 (25%) (p = 1.00)
18 (32%) (p = 1.00)
Head and neck
22 (19%)
13 (20%) (p = 1.00)
15 (26%) (p = 1.00)
45 (38%)
24 (37%) (p = 1.00)
24 (42%) (p = 1.00)
Scar location
Number of participants with restriction in range of motion or joint contractures or scars pulling on other body parts due to scarring Grafted
73 (62%)
45 (69%) (p = 1.00)
41 (72%) (p = 1.00)
3 (3%)
1 (2%)
0 (0%)
1 (1), 0–16
1 (1), 0–16 (p = 1.00)
1 (2), 0–16 (p = 1.00)
3 (3%)
1 (2%)
0 (%)
Median age in years (interquartile range), range
34 (21), 18–85
34 (26), 18–80 (p = .32)
34 (22), 19–80 (p = 1.00)
Median Fitzpatrick skin type (interquartile range), range
3 (3–4), 1–5
3 (3–4), 1–5 (p = 0.32)
4 (3–4), 1–5 (p = 1.00)
1 (1%)
0 (0%)
0 (0%)
4 (1–10),
